THE PRE AND POST OF HARDWARE IN THE LOOP EXPERIMENTS FOR POWER ELECTRONIC POWER DISTRIBUTION SYSTEMS
Abstract
Thanks to the Power Electronics Building Blocks (PEBB) concept there has been significant progress in the definition and design of Power Electronics Power Distribution Systems (PEPDS). Recent developments in SiC power devices are yielding PEBBs with far greater switching frequencies than Si based devices, resulting in an order of magnitude reduction of the time scales that characterize converters operation, if compared to systems utilizing conventional IGBT basedPEBBs. Recent development on the FPGA based Universal Controllers significantly increased the computational power available at all layers of PEBB control, from gate drive circuits to system level coordination. The increased computational power, together with high-speed communication technology, enabled the development of Networked Universal Controllers that allow the application control layer of converters to serve as the most fundamental system control layer. Their coordination across the system ~using appropriate communication architectures~provides a degree of energy management not previously realizable in power systems; potentially improving flexibility, efficiency, reliability and survivability of ship systems. At the same time the distribution of functions and the high level of interoperability of PEPDS make their designand operation ~so that the mentioned advantages are really achieved~ a challenge. The complexity of PEPDS design and operation is also accentuated by behaviors and dynamics that are more and more determined by software.Hardware In the Loop (HIL) and Power Hardware In the Loop (PHIL) are widely considered the state of the art in energy systems testing and in the last twenty years significant work has been performed to analyze and improve the accuracy and stability of the needed interfaces. In those years, HIL and PHIL have been mainly used as fast prototyping tools and more recently to verify the operation of the DUT as alternative to traditional laboratory and field testing. The main objective of this project is to define data driven modeling techniques that allowcapitalizing on the results of HIL and PHIL testing by creating models of the devices under test ~also for closed-source, proprietary systems~ using the collected data. The model so obtained can be used ~for example with tools like Smart Ship Systems Design (S3D)~ at various stages of a PEPDS design but they can also be used during PEPDS operation as part of model based controlscheme as well as digital twin of the PEPDS. To verify the operation of the different components of a PEPDS system is important that the HIL and PHIL experiments are systematically defined. The second objective of this project is to define a stochastic experiment design approach for HIL and PHIL testing. The two objectives are strictly related since the experiment design will also determine the range of validity of the models developed for the Device Under Test (DUT). For the development of data-driven models we plan to use Artificial Neural Networks (ANN) to populate pre-defined model structures. Preliminary work, using this approach, demonstrated promising results in the modeling of a grid-connected inverter. For the design of HIL and PHIL experiments we plan to use a stochastic approach based on non-intrusive Polynomial Chaos (PC)levering the previously performed work.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 07, 2019
- Source ID
- N000141912060
Entities
People
- Andrea Benigni
Organizations
- Office of Naval Research
- United States Navy
- University of South Carolina