Anomaly Detection Toolkit for Integrated Systems HealthManagement (ISHM)

摘要

THEgoal of Integrated Systems Health Management (ISHM) is to provide measures of system health using THE data combined with models of behavior. Many technologies contribute to creating an ISHM capability. Arguably the most important of these are methods for anomaly detection, because that is the source of events that trigger downstream reasoning about cause-effect relationships. This paper describes recent efforts to create an anomaly detection toolkit to support on-going ISHM development at NASA-SSC. The toolkit consists of two major components. A health assessment database system (HADS) based on MySQL implements a repository of test data and related configuration data; a companion browser tool allows access to the HADS test data, supports queries, and allows the data to be analyzed using available anomaly detection algorithms. The browser provides a graphical user interface to make the tool usable by operations personnel.