Introduction to Mr.Chen Linchevski
Chen has vast experience in management, sales, and business development of industrial software. Chen is a true believer in the ability of the industrial sector to transform itself and increase efficiency as never before, by using Big Data.
Previously, Chen was the CEO of Opcat, in leading positions in Ayeca and worked with key technology companies such as GE, Israel Electricity Company, Elbit, and IAI, specializing in analysis and design of complex business analysis. Chen holds LLB from the Hebrew University in Jerusalem.
Sharmila Annaswamy (SA), Senior Research Analyst for Industrial Automation & Process Control Group had an opportunity to interact with Mr.Chen Linchevski, CEO & CO founder of Precognize for a Movers and Shakers interview.
SA : Can you start by providing our readers a brief overview of Precognize, including the vision behind its formation and its current role in the market?
CL: Precognize is a data analysis and equipment failure prediction software for asset heavy industries such as chemical, oil and gas, petrochemical, and steel making. Right now we have partnered with GE Predix to develop applications on their Predix platform for predictive analytics applications for Predix cutomers.
SA: Can you describe Precognize for our readers? What is your customer value proposition?
CL : Precognize brings customer value by predicting failures of sensors, pumps, heat exchanges, or any other process equipment that might seem to fail suddenly and enable customers to prevent its occurrence. As asset heavy regulated industries such as chemicals and oil and gas already have an installed base of sensors and historian data collected over a period of years, many customers are eager to derive value from that data and Precognize offers exactly that.
SA: What is the unique value proposition of Precognize and what are your key competitive differentiators?
CL : The key differentiator about Precognize is that it provides predictive analytics for the entire plant and not for a component such as pumps and compressors. Another key differentiator is that we eliminate false alarms; Precognize removes the major implementation barrier by providing its users with authentic and actionable alarms in place of multiple false alarms. Precognize also offers the user the flexibility to track, analyze, and infer any form of information that is required such as identifying operator mistake, leakage assessing mistakes and so on. Precognize also provides quick implementation of 2 weeks to the customer.
SA : Can you elaborate on the acceptance or adoption of Precognize in the marketplace ?
CL: Leading chemical producers such as BASF and leading oil and gas producers have utilized Precognize to derive value from their data. In real time, Precognize was able to identify true events and generate true alarms for 99% of the time. This hit rate would enable operators to plan plant shutdowns at an earlier stage and implement maintenance activities to prevent future failure. This prevents any unplanned sudden shutdowns. Over time the model will move from planned shutdown to complete prevention of shutdown modes. As a plant shutdown of even an hour could potentially result in loss of millions of dollars, organizations are increasingly interested in investing a small amount which could go a long way in preventing future shutdowns.
SA: Can you elaborate on the technology differentiation and how exactly does Precognize eliminate false alarms?
CL: In traditional predictive algorithms, the system triggers an alarm each time a control system crosses an upper or lower threshold, leading to multiple alarms that actually do not require operator attention. Other models utilize historical data to predict specific types of problems, but the problem is that there are not many historical events to compare to for specific prediction. Another possible approach is to aggregate data from various customers and to use that pool of information for anomaly detection.
Precognize understands that due to the presence of innumerable sensors and clusters and the sensitivity of the operation, many of the anomalies detected will be irrelevant to the customer. Precognize’s machine learning model is ingested with historical data where the data are analyzed in engine by unsupervised learning to reveal clusters of normal operation. These are then used as base data to identify abnormal clusters during production. The engine also does in-depth analysis amongst systems and subsystems to identify the malfunctioning element.
SA : How does Precognize exactly classify an anomaly as true/ false? How does Precognize engage with customer to achieve desirable results?
CL: Along with historical data, plant behavioral data are fed in to the system in order to model it as close to real system. The data about the machine operation process are fed using local ERP and SAP systems. Apart from this, the actual operation knowledge which lies with the operator is digitally transformed in Precognize software. The operator is required to model the plant by hierarchically identifying and ordering the structure/sub structure process/ process divisions/ until the component level and map what information is being measured straight on the Precognize platform. Using this Precognize maps the component physical data to the process data and maps the critical components accordingly. This is done by the operator themselves during the implementation period. By using this approach we prevent information mishap and ensure the plant is modeled by the field experts. This is a step toward facilitating self-service analytics for plant technicians without depending on data scientists. This also helps in eliminating noise and performing operations without any doubt on third-party service provider.
SA: What is the growth strategy of Precognize to sustain innovation in the future?
CL: Partnerships are the key growth enablers and we have partnered with platform providers such as SAP, GE, Azure, Siemens to ensure that we reach out to all end-user industries. We also partner with consulting companies who advise digitization and strategy changes for IoT implementation to their clients. Another major partnership we are looking at is large conglomerates that are outsourcing their maintenance needs. Prescriptive analytics is going to be the natural growth extension of Precog, but the only problem is that prescribing exact code of action is going to be difficult for process industries. As Precognize software platform is capable of running independently and completely without our intervention, we take care of the distribution and updating part of it only.
