Anomalies and prefailure states are inherent and often unavoidable working conditions of industrial installations. Undetected in time, it leads to failure. Early detection of such symptoms avoids failures and reduce unplanned plant outages. It also enables the optimal planning plant maintenance. This reduces the costs of maintenance and repairs, increases equipment reliability and plant availability and improves safety. The Exergon company has developed AURORA software which based on artificial intelligence methods to detect anomalies and prefailure states in industrial plants. The data used by the AURORA software are the standard measurement data collected at the plant. In most cases, the implementation of the anomaly detection does not require increase the number of measuring devices. The AURORA system works in real time to inform the technical staff of the states deviating from normal operation, which in the near future may lead to failure. What’s more, the AURORA software identifies possible cause of the anomaly allowing you to take early corrective action and avoid failure. Installed system does not require continuous monitoring of data by the technical staff in the plant. Warnings are sent only when the system identifies a pre-failure conditions. The system architecture is provided so as to minimize the need to change the configuration of the IT infrastructure at the plant and use already installed software for data collection. This significantly reduces the costs of introducing the developed anomaly detection system.