ReNew Power is one of the largest renewable energy Independent Power Producers in India and globally. ReNew Power develops, builds, owns, and operates utility-scale wind and solar energy projects, and hydro projects. As of June 14, 2022, ReNew Power had a gross total portfolio of ~12.8 GW of renewable energy projects across India.

Problem Statement :
Unplanned downtime of wind turbines can result in a significant loss of revenue and energy and can easily scale to millions of dollars a year. It is therefore pivotal that flagging of the failure of components is made to prevent further loss and perform maintenance. It, however, involves replacement of components and higher costs. Condition-based monitoring systems rely on supervisory control and data acquisition systems to predict faults and get valuable insights into the turbine’s performance.

ReNew Power shared minute-wise normalised data of wind speed, power and temperature data for multiple components of a wind turbine. The company is looking to create a model to get an ideally functioning turbine’s expected rotor bearing temperature. It will then use the model to check the deviation of the actual rotor bearing temperature of the faulty turbine from the expected temperature.

Evaluation :
The submission is evaluated using the Mean Absolute Percentage Error.

Data
https://machinehack.com/hackathons/renew_power_hiring_hackathon/data

NOTE: “Timestamp is the unique identifier of the data; it shouldn’t be used as an input in the model”

After applying various algorithms like Linear Regression, Lasso, Ridge, Random Forest and Xgboost and Xgboost gave the best performance and my MAPE score comes out to be 0.01925