Grid4C

Decentralization of energy supply and deregulation of energy markets, coinciding with exponential increases in the data available from energy users, have created a pressing need for software to understand energy consumers and manage energy resources. Underlying each of those requirements, are challenges that cannot be met by the analytical approaches that have been employed in the past. Predictive Analytics that forecast behavior and faults of individual meters and connected devices meet these challenges with a product suite built on a core set of patented Machine Learning algorithms.

"Grid4C’s edge lies in the ability to squeeze the greatest value from existing, ubiquitous data sources, non-intrusively, without needing to wait for new sensors to reach mass adoption"

"Machine Learning provides a window into homes and businesses using smart meter data that facilitate granular, accurate predictions, enabling more reliable operation of the grid and integration of renewable resources and energy storage,’’ explains Dr. Noa Ruschin- Rimini, Founder and CEO, Grid4C. Over the years, the company has been able to solve some of the industry’s greatest challenges, outperforming the competition in industry benchmarks. Having launched a partnership in APAC with smart grid and smart metering leader Landis+Gyr, Grid4C's software today is deployed for more than one million homes and businesses and analyzes billions of meter reads from four different continents, producing more than 500 million predictions daily in less than an hour.

Energy Providers Adopt Cloud Computing

One of the big changes in the landscape is that utilities recognize that cloud computing is not only an option, but a necessity if they want to take advantage of machine learning and advanced analytics.

Grid4C’s solutions help this market on three fronts: customer-facing applications that help businesses and consumers not only save money but predict problems with the appliances they rely on, predictive customer analytics that facilitate segmentation and micro-targeting and predictive operational analytics that optimize procurement, grid operations and the integration of solar, energy storage and electric vehicles. "Customer demands are just starting to catch up with the capabilities that we can provide," says Dr. Ruschin- Rimini. A solution that is generating a lot of excitement in the market is Fault Prediction, Detection and Diagnostics (FDD) based on smart meter data, which can be enriched with data from smart appliances such as connected thermostats, enabling the same algorithms to deliver deeper diagnostics and insights.


As an example, ‘‘with smart meter data, we can not only detect mechanical problems and wasteful settings but can predict problems before they happen,’’ explains Dr. Ruschin-Rimini. This can prevent slight inefficiencies from transcending into larger problems of much more severe consequences. By utilizing a highly modular architecture, Grid4C can easily implement and customize the core analysis and insights derived in the process. ‘‘The core of our products is our proprietary AI self-learning engine, so all you need to do is ‘throw’ any data that may be relevant into it,’’ claims Dr. Ruschin-Rimini.

Benefits of a Machine Learning System

Machine-Learning holds the key for generating new revenue streams, optimizing operational processes, for better balancing the grid, reducing spikes during peak hours, optimizing Demand Response and pricing programs, identifying appliances that are about to break, pinpoint costly behaviors that are easily fixed, and essentially understand homes, businesses and their occupants.

"By monitoring each meter separately and analyzing data like meter and device reads, weather data, customer data and more, Grid4C's engine automatically learns the underlying correlations and hidden patterns and generates predictions in a plug-and-play manner," explains Dr. Ruschin-Rimini. The plug-and-play approach allows predictions to be generated very quickly, which means customers have the added benefit of short time-to-value.

By using information theory-based algorithms, the machine learning engine can decompose the behavior of each meter into sub-series and, combining the analysis of each meter’s data with customer data, develop a deep and detailed understanding of customers’ homes and businesses from both physical and behavioral perspectives.

Behind The Machine

Grid4C’s success stems from its strength in Machine Learning and data science.