Anand Rinchinjugder

Practical data science project focusing on real-world applications and techniques.

Substations are easily susceptible to failure and the cost of such a failure is substantial. For instance, a severe ice storm in China in 2008 caused widespread power outages that imposed an economic loss exceeding 2.2 billion USD on the country, prompting a reassessment of its electrical grid infrastructure in response to the snow and ice storms encountered that year. There are no comprehensive real solutions to address this vulnerability effectively. Short term outages, referred to as momentary outages, typically have causes that are not recorded in outage datasets. However, repeated momentary outages can escalate into long term permanent outages, thereby substantially diminishing overall system reliability. This proposal focuses on predicting substation failures using incomplete, noisy, and sensitive data from electrical substations using modern machine learning techniques.