Startup hopes to halt pipeline leaks by predicting threats
Israel’s Precognize uses artificial intelligence to detect glitches as they are born and send out clear alerts
In September, Minister of National Infrastructure, Energy and Water Resources Yuval Steinitz was forced to temporarily authorize the use of alternative fuel sources to produce electricity, as a cracked pipe in Tamar — Israel’s only commercial natural gas field — shut down production for nearly a week.
It is problems like this that Precognize works to prevent, using artificial intelligence to analyze data, detect risks and send out warnings, said CEO and founder Chen Linchevsky.
“In many of the cases we’ll be able to tell you that something is starting to fail,” he said.
Using vast amounts of data received from thousands of sensors installed around utilities or manufacturing plants, the predictive maintenance technology developed by Precognize can translate the raw information into specific warnings in time to prevent problems and prepare a response.
Founded in 2011, the Tirat Carmel, Israel-based startup has managed to secure customers internationally, including the Germany’s BASF — the largest chemicals producer in the world — and the Israel Electric Corporation.
Linchevsky said industrial security systems collect massive amounts of data, but current technology lacks the ability to automatically transform that data into concrete early warnings that would reduce or altogether prevent any damage. And herein lies what Precognize considers its key innovation and market advantage.
“In a very complex environment there are thousands of [data] anomalies, and there is a huge gap between anomalies in the data and real problems in the plant,” said Linchevsky. The irregularities don’t generally appear in a visual form and often it is difficult to understand if there is a real issue that needs attention or if it is a false alarm, he said.
“We are one of the few companies who have bridged this gap,” he said.
Precognize’s technology builds a model representation of the plant and projects the data anomalies onto the model “in order to aggregate all this noise into a few meaningful alerts,” said Linchevski. He attributed this ability to his background and that of other Precognize executives in conceptual design and system engineering at the Technion — Israel Institute of Technology.
In a typical large-scale industrial plant with thousands of inter-correlated sensors, just data science “isn’t enough,” said Linchevski. Combining data analysis with system engineering, the anomalies no longer appear as random fluctuations of data but as material situations that affect the functioning of the plant, using a visual interface that Precognize claims “allows operational experts with no modeling training to accurately describe their systems.”
Precognize innovation stems from mixing the two disciplines, data analysis and conceptual design, said Linchevski. Because so few companies are doing this, the CEO said, Precognize doesn’t see much competition at the moment.
Precognize got its first big break when Israel Electric allowed it to test its beta technology on plant equipment. Shortly after that, Precognize was chosen out of 300 possible competitors by German giant BASF to install its predictive maintenance technology in its plants, and from there things “started rolling,” said Linchevski.
In late 2016 the startup secured some $2 million venture capital funding from Maverick TLV, allowing the startup to put its bootstrap days behind it, Linchevski said. Now it’s looking to expand its technology to other fields, such as transportation.
