Abstract: An apparatus and method for detecting and classifying in real-time characteristic of a system component is provided. A sensor senses the system component and outputs a first quantity of data corresponding to a characteristic of the system component. A modeler receives the first quantity of data, converts it to a numerical value and runs a computer model simulation to detect an anomalous behavior of the system component. The detected anomalous behavior is optimized and expressed as a second quantity of data. An autonomous reasoner collects the second quantity of data. A database has a plurality of signatures related to predominant modes of the system component. The autonomous reasoner compares the second quantity of data with the signatures and identifies a signature that matches the second quantity of data. An output indicative of a cause of the anomalous behavior of the system component is provided.