April 2017 Digital Edition
March 2017 Digital Edition
Feb. 2017 Digital Edition
January 2017 Digital Edition
Nov/Dec 2016 Digital Edition
Oct 2016 Digital Edition
BRS Labs shows AISight behavioral-recognition-based analytics at ISC West
Las Vegas, NV, April 21 – Artificial intelligence, which has been used to entertain through toys such as Tamagotchi, as well as to reap profit from picking investments, is now being used by Houston-based Behavioral Recognition Systems Inc. (BRS Labs) to teach itself typical patterns in surveillance analytics and to send alerts that may help companies avert disaster.
The company’s “Behavioral Recognition Without Human Interference” capability is increasingly being successfully used at airports and seaports, utilities, in public safety and security and in critical infrastructure protection.
Based on a technology that allows computers to learn when operations are normal, the system calls out abnormal activity. This ability can be applied to different sensors, industries, uses, temperatures and flows, and has been expanded within the last year to SCADA, according to Wes Cobb, chief science officer, at the company’s booth at the 2015 Las Vegas trade show.
Users do not tell BRS Labs’ newest technology what they are looking for. And users do not need to become experts in conducting analysis. The system, based on BRS Labs’ signature AISight technology, tells its users what is unusual, based on a typical view from one or multiple cameras, plus other data.
At ISC West, BRS Labs shared how its trademarked AISight, an artificial intelligence-based system, compares tremendous volumes of data gathered from hundreds, even thousands of cameras, as well as other inputs. From this data, the technology determines what is typical – and sends a proactive alert on what is out of the ordinary.
“The analytics, which were based in video, were applied to multiple sensors and now are capable of SCADA control analysis, ultimately, transforming beyond video and gaining acoustic information and other information sectors,” said John Frazzini, president of the company.
“The BRS Labs artificial intelligence system had been refactored so that it can compare data of many different types, learning as it goes through the new artificial intelligence learning modules. We refactored the architecture of our learning engine for new sensors,” Cobb explained.
Besides video stream and acoustic information, BRS’ latest machine learning system takes data such as electrical current, electrical flow and temperature into account. The analytic system teaches itself the normal patterns for these various data – and is capable of recognizing unusual patterns across several sensors at the same time.
“The machine learning engine accepts data and builds a dynamic language to create and learn what is normal, and to describe that in a particular language,” Cobb said. This occurs over 30,000 to 40,000 data points, essentially single facts of information, he said, “enough to obtain plausible statistics.”
These statistics are based upon behavior, such as occurrences during a certain time of day, a certain day of the week. While comparisons, machine learning and AI have been used previously to examine other types of information, Cobb said, “video has taken a longer time to reach maturity.”
According to Jim Thompson, BRS Labs new CEO, creating a mash-up of data is not the ultimate point of this technology. The point of the system integrating data from many different sources is to alert users of anomalies. “There is no apparent reason why more sophisticated, automated alarms and algorithms cannot be built into a display system to alert when anomalies arise,” said Thompson. “In light of the potential consequences, it is no longer acceptable to rely on a system that requires the right person to be looking at the right data at the right time, and then to understand its significance in spite of simultaneous activities and other monitoring responsibilities.”
With this type of monitoring, Thompson pointed out, “the return on investment is in days, weeks and months, not in years.” Because of how the system operates, users can expect an alert in the case of unusual activity, before a SCADA alarm ever signals. “
For instance, a steam sensor in a refinery system might be on the verge of malfunction, unrecognized by human operators. The AI technology-based system could detect a problem a week in advance of this mechanical failure that could mean shutting down the refinery. In another instance, an AI system, realizing the system flow and rate on a midstream pipeline, produced an alert for a gas leak that the SCADA system had not yet recognized. In this instance, the line was losing pressure – a signal that might impact life and safety issue.
“AISight, for SCADA, is like having 50 operators helping me watch what’s going on all the time in the system,” Thompson said – then sending alerts on “things I don’t know.”
A BRS press release about the system calls the real-time video alert a “tap on the shoulder” for security personnel, providing the means for them to mitigate risk and damage.