Data-driven system for predicting outages and their severity

点击翻译按钮获取中文摘要

Advances in energy generation processes as well as in customers' needs and expectations lead to fast changes and modifications of a power grid. Variability and uncertainties in a power distribution grid are observed on the generation side due to growing involvement of clean energy sources and potential contributions of small, independent energy producers; and on the load side due to changes in profiles of energy consumption, nature of devices connected to a grid, and increased presence of energy storages. Improvements in monitoring and data collection practices provide opportunities to more comprehensive modelling and managing operations of a grid. A power grid can be enhanced and become more efficient by integrating techniques of artificial intelligence in analysis and optimization of its operations. Advanced data analysis methods should be able to address a service quality degradation due to outages, weather patterns and asset related performance.The proposed project aims at applying Machine Learning/Data Mining and Computational Intelligence methods for analysis of power distribution system data, and developing a methodology for processing this data. The goal is to construct a system for predicting power outages and their severity. Weather data and description of system's topology will be utilized by the proposed system for predicting purposes.It is expected that the developed prediction system will lead to identifying probability of outage occurrences as well as their impact on customers and the power system. Additionally, it is anticipated that the work perform during this project will lead to discovery of new insights aiming at augmenting and predicting reliability of a power distribution grid.