Energy Analytics

Problem Statement: Reducing maintenance energy expenses on cell sites

Whiteklay works with one of the biggest cell site service provider to help them gain efficiencyon energy using Izac Framework, with the below objectives :

  • Realtime ingestion of all IOT data from cell sites into Izac platform
  • Prediction of Generator Energy
  • Prediction of Mains Energy Consumption
  • Prediction of Total Energy Consumption
  • Prediction of Fuel Consumption
  • Analytics on Data for various business benefits

solution

To create a single realtime analytical view of the cell sites, solution needs to collect,process, store and analyze huge data in that too in real time. We proposed Izac dataexchange system which provided them with a consistent, standardized way ofexchanging data and information. The data exchange platform is built using open sourcetechnologies and will provide a consistent data repository which can help universities inmanaging and sharing their data assets. It provides features for effective collaborationwithout compromising data integrity and access controls. The tool is built using thestandards like mentioned in https://www.internationaldataspaces.org/ which lay downthe principles of data sharing. Some of the features of the tool are as follows:-

  • Realtime data connectivity to multiple sources of data
  • Inbuilt Data Profiling capabilities to check the quality and structure of data
  • Data Marketplace for effective collaboration.
  • Rich Functions to transform data assets
  • Significant reduction in data engineering and can be easily used by non conders
  • Metering data usage for individuals to check data usage
  • Easy and standard way of sharing data to industries or other academic institutions for research and development
  • Mitigate legal risk by adding consent management to the data assets
  • Auditing of data assets

The Solution consists of following phases:

  • Data Extraction and Ingestion
  • Data Cleaning and Analysis
  • Model training & Prediction
  • Visualization

Data cleaning and analysis

  • Fitering of Unrequited noise and disturbances from sensor data for better analysis.
  • Analysis of cleaned data for model selection, insight generation and primary testing of accuracy.

Model Training & Prediction

After model selection and creation of algorithm, best ensemble machine learning model was trained against the data and predictions were generated for each sight.