NASSCOM Community

Analytics Case Study Series: 2. Energy & Utilities

Blog Post created by NASSCOM Community on Jun 9, 2016

The Utility industry can be brought to the internet age through Automated Meter Infrastructure (AMI) and Smart Grid. Enabling direct communication with metering devices opens up a realm of possibilities for understanding usage patterns that were not even registered. The ability to measure and analyze data about energy distribution and consumption on a more granular basis—in time and in detail—can unlock significant value.

Key Application Areas:

 

  • Demand response
  • Revenue management
  • Fraud and loss prevention
  • Energy efficiency
  • Compliance
  • Asset maintenance and management
  • Customer care and management
  • Forecasting and load management

 

The Case Details:

 

Organisation: Business Brio

Client: An energy company

Vertical: Wind Energy

Geography: The Netherlands

Category: Predictive

 

The Situation:

 

  • Challenges in accurate prediction of wind energy generation at a particular point of time, due to very stochastic nature of wind energy generation

 

  • This prediction was critical to save conventional energy due to 8/12 hour lag in ramp-up/down of conventional plants

 

The Solution:

 

•Regression and time series method to predict wind energy generation in shorter time horizon (<= 8 hrs)

 

•Artificial neural network and ensemble with (ordinary linear regression) to forecast for medium time horizon (>8 and <24 hrs)

Results:

 

Forecasted wind energy supply in a power grid for understanding the need of conventional energy as

  controlled input with 84% and 76% accuracy in cases of short term and medium term prediction respectively. 

Outcomes