05/07/2018

Press release: Production Cybersecurity Project Is Implemented at JSC TANECO

They started pilot operation of an early detection system for deviations (anomalies) in the technological process of the CDU / VDU-7 unit operation at the TATNEFT’s TANECO Refinery Complex.

The operation stability of CDU / VDU-7 is critical for yielding quality raw materials used for further processing at the refinery. Technological processes taking place at the unit are sensitive to such parameters as input raw materials; correct functioning of the cooling system, as well as the operating temperature range. Therefore, it is very important to receive prompt information about any deviations from the normal operating process in the automatic mode.

They have selected MLAD (Machine Learning for Anomaly Detection) technology, proposed by JSC Kaspersky Lab to ensure the safety of the technological processes at the CDU / VDU-7 unit. The companies have been cooperating for many years, and they began working together in October of last year on a new project that was implemented with application of the machine learning technology for the telemetry of the technological processes and artificial intelligence to detect cyberattacks on the CDU / VDU-7 installation. They developed a neural network model in 2017 capable of detecting anomalies. If any previously unaccounted data on the normal system’s operation become available, then the system is capable to get additional learning. All this resulted in taking a defensive solution based on machine learning algorithms possessing larger flexibility, unlike an expert system that operates on a set of rigidly defined rules.

The pilot system operation in real-time monitoring mode with automatic detection of process deviations from their normal behavior was started in February 2018. During the work, they succeeded to detect various types of anomalies: deviations of the technological process associated with the regime change periods; switch over of the control loops to manual mode; situations associated with incorrect sensor readings.

Thus, the system introduction allowed the information security and technological processes specialists receiving a tool for automatic early warning of dangerous situations, detecting the anomalies and their interpretation, as well as an intuitive interface with trends in process parameters and analysis of deviations.


Другие новости этого раздела: