Risk Edge forays into Renewable Energy Sector with AI offerings

(Mar 2018) Risk Edge has forayed into Renewable Energy segment with focus on predicting Machine Failures for Wind Turbines. This initiative coincided with Risk Edge being selected by one of India’s largest Renewable Energy players for predicting Machine Failures for their Wind Turbines using SCADA data.

Speaking on the development, Nitin Gupta, CEO of Risk Edge said, “Globally, the wind energy sector loses about USD 2.5 bn due to Wind Turbine Failures. If the operational costs are also taken into consideration, these losses can reach upto USD 7.5 bn. It is thus imperative for the Wind Energy players to focus on limiting such losses to improve operational efficiency and ROE. With recent advances in computing power and Machine Learning Algorithms, it is worthwhile to invest in technology to predict machine failures in time and reduce these losses. As there is not much research available in this domain (compared to other areas like ecommerce / financial services), Risk Edge aims to be a pioneer in this area by evolving custom-built algos for predicting machine failures.”

As part of the project, Risk Edge will be given access to the entire historical SCADA data for many Wind Turbines of the company. Risk Edge will deploy it Machine Learning experts, including PhDs in AI to develop algorithms to predict machine failures. Speaking about the complexity of such projects, Mr. Gupta said, “Since not all wind turbines are from the same OEM vendor, the SCADA data differs to some degree from turbine to turbine. So we’ll first have to normalize the data across wind turbines from different vendors. Besides, selecting features from over a 1000 different features and then do feature engineering, also requires domain knowledge which our team has acquired by talking to domain experts in this area. This domain also has many dimensions across which machine learning can be applied to answer different questions. From selecting the right data frequency, to analysing machines based on their age & maintenance records, clubbing turbines located in different geographical regions to ensure similar weather conditions for them, each opens up a wide array of dimensions across which machine learning can be applied.”

On handling large amounts of data, Mr. Gupta said, “Risk Edge has done some good work in managing Big Data projects earlier, and we intend to bring our learnings from there into this project. The approach we are taking here is quite simple – Think Big so we know the possibilities that come with dealing with such data, build small prototypes so as not to get mired in technology at the very beginning of the project, and then Scale Fast as we achieve a higher level of accuracy with our customized algorithms.”