JOHNSON Controls has released a false alarm reduction service that applies machine learning algorithms to intrusion alarm panel data and provides end customers with insights to prioritise actions said to vastly reduce preventable nuisance alarms, according to an SSI report.
To train its algorithms and confirm their accuracy, the company crunched 5 years of numbers from the 757,000 panels it has in the field. An 18-month field test of the resulting algorithm lead to a 50-70 per cent reduction in client false alarms.
The algorithms have been proven to be 96 per cent accurate, Christine Trainor, director of data intelligence at JCI and co-inventor of the solution, told SSI.
“This is our 1.0 release. We are not in the monitoring center on the back end of the receiver. We have a roadmap to get there, but our initial release is to all of the preventable alarm activity that really the customer owns and can prevent by decisions being made on the business side,” Trainor said. “It is about eliminating that alarm activity altogether.”
No physical deployment is needed, and the software system is panel agnostic. The only requirement is that the customer use the MASterMind integrated software platform by United Technologies Corp, which processes data to uncover recurring patterns that are mined to provide customised recommendations to reduce false alarms.
“On our journey we discovered an incredible amount of manual effort that is put in place to try to stop the chain of events to dispatch,” said Trainor, who invented the machine learning solution along with JCI data scientist, Gopi Subramanian. “What we’ve done in this solution is to say, ‘Let’s first tackle the problem on the front end of the buffer to remove as much as possible.’”
According to Trainor, the JCI solution could have eliminated 5800 of one client’s false alarm signals from ever triggering to the receiver. With an estimated US$3.2 billion spent annually on false alarms fines and associated costs by organizations and municipalities across the U.S. the JCI solution offers big savings.
“Most of these corporations don’t realize what their cost is. A lot of fine activity is hidden at a local level. They put it in where they pay taxes because they will get in trouble if it goes to corporate,” she said.
The subscription-based service is a new topline growth revenue target born out of JCI’s recently established Data-Enabled Business. The solution is agnostic and can run on any Cloud computing plat-form, such as Microsoft Azure or Amazon Web Services.
It can operate as an on-premises solution in which case JCI will license the software to the customer. Alternatively, strategic partners can license the solution and operate it as a proprietary cloud service.