Patented Fast Data Analytics

PowerWise Patents Fast
Machine-Learning Algorithm

PowerWise, through its Chief Technology Officer, Curtis Meadow, received United States Patent 10,489,716, Method for performing automated analysis of sensor data time series, at the end of 2019.

Almost every business generates, manages, and analyzes an incredible amount of data to make decisions. Predictive analytics are often computationally expensive and human-resource intensive as well. The new patented method, known within PowerWise as Fast Data Analysis and Response (FADAR), is a self-learning system based on a modified data analysis algorithm called K-means clustering. The modified algorithm is used to find an optimal solution for the value K that accurately represents the operating states of equipment, using a variety of sensors as raw input data. A carefully-designed sequence of iterations on large amounts of live streaming data provides an exceptionally efficient method to determine complex patterns. Meadow estimates that FADAR is several orders of magnitude (10 x 10 x 10 x 10...) faster than conventional methods of identifying complex patterns in big data, as illustrated below.

FADAR is lightning-fast at recognizing complex patterns

The technical steps in the process, simplified for illustrative purposes, comprise two phases: 1) a brief learning period and 2) straight to active service. The 30-day learning period may be significantly shorter, depending on data complexity.

The ultimate goal of cluster analysis in FADAR is to find a set of information that corresponds to a normal operating state of a complex piece of equipment.

Such pattern recognition currently requires entire datasets to be exported for analysis, and the analysis may take weeks. FADAR learns from incoming data over a period of about 30 days, analyzing the data clusters during the server’s quiet periods to determine normal operating states. After the learning period, it can recognize deviations from normal operation in real-time as data comes in, a process tens of thousands of times faster than current analytical methods.

Read About Use Cases for Fast Machine Learning »

PowerWise develops and sells energy and building management systems to help people reduce costs, better manage their facilities, and achieve energy and operational efficiency. The company is based in Maine. For more information, visit www.powerwisesystems.com or call +1-207-370-6517.