top of page
  • safwanwshah4

Grant awarded

In collaboration with university of Rochester I have been awarded a grant from NYSERDA, to research advanced machine learning algorithms for parameter verification and calibration

University of Vermont in partnership with Rochester Data Science Consortium scientists at the University of Rochester, and the New York Power Authority, have been awarded $225,000 from the New York State Energy Research and Development Authority (NYSERDA) to apply the latest machine learning methods to develop accurate, reliable, and scalable algorithms for modeling electric generator power systems.

If successful, this “proof of concept” research has the potential to lay the groundwork for modernizing today’s modeling methodology for power systems.

“Having accurate system models and scalable modeling tools are indispensable to the power system’s overall safety and reliability,” says Beilei Xu, principal investigator and senior research scientist at the Rochester Data Science Consortium. “By introducing machine-learning based methods to current modeling tools, this research has the potential to provide more sophisticated modeling methods for the smart grid.”

Traditionally, the most common method for performing power plant model validation is through offline staged tests, which require the generators to run and produce greenhouse gas emissions even though the power is not being fed into the grid. Further, the cost of this method is considerable and companies lose revenue when the power plants are offline. For agencies and departments that own a large number of power plants that need to be tested repeatedly, the traditional method makes a significant impact on their bottom line.

Recently, online optimization methods have used Phasor Measurement Unit, an advanced sensor technology, to collect data, enabling significant progress in updating the old processes. However, they still suffer from being unreliable and difficult to scale.

The NYSERDA-sponsored research includes the performance evaluation of two synergistic machine learning approaches – deep learning and reinforcement learning – for model validation. These approaches can potentially enable testing to occur during normal operations, which improves the scalability and efficiency of the model validation process. This is the direction that model methodologies need to be headed, if they are to keep up with the growing complexity and sensitivity of the power grid.

NYSERDA's smart grid program focuses on advancing New York State's goals to modernize New York's electric grid by integrating new technologies to support Governor Andrew M. Cuomo’s nation-leading mandate for 70 percent renewable electricity by 2030. This project also supports New York’s nation-leading target to reduce greenhouse gas emissions 85 percent by 2050.

“Developing and bringing innovative smart grid technologies to the marketplace is a key component to modernizing our electric grid so we can provide New Yorkers with a more resilient and reliable energy system,” says David Crudele, program manager of Smart Grid Systems and Distributed Energy Integration at NYSERDA, “This project has the potential to make a significant contribution in Governor Cuomo’s fight against climate change by reducing greenhouse gas emissions from power plants during testing and I look forward to watching it progress.”

By using advanced machine learning algorithms applied to the PMU data, “a larger scale of power gird assets can be effectively and reliably verified and thus help New York's Clean Energy plans of being 100 per cent carbon-free by 2040,” says Safwan Wshah, assistant professor of computer science at the University of Vermont,

Ramadan Elmoudi, research and technology development engineer with the New York Power Authority, says, “Power system modeling is crucial to perform power studies and evaluate the impact of different power system contingencies in order to avoid unscheduled outages and increase power system reliability, resiliency and availability; therefore, implementing cutting edge technologies will help the power system stakeholders achieve these goals.”



bottom of page