{"@type": "dcat:Dataset", "accessLevel": "public", "accrualPeriodicity": "irregular", "bureauCode": ["026:00"], "contactPoint": {"@type": "vcard:Contact", "fn": "Kai Goebel", "hasEmail": "mailto:kai.goebel@nasa.gov"}, "description": "For the purposes of this paper, the National\r\nAirspace System (NAS) encompasses the operations of all\r\naircraft which are subject to air traffic control procedures.\r\nThe NAS is a highly complex dynamic system that is\r\nsensitive to aeronautical decision-making and risk management\r\nskills. In order to ensure a healthy system with safe\r\nflights a systematic approach to anomaly detection is very\r\nimportant when evaluating a given set of circumstances\r\nand for determination of the best possible course of action.\r\nGiven the fact that the NAS is a vast and loosely integrated\r\nnetwork of systems, it requires improved safety assurance\r\ncapabilities to maintain an extremely low accident rate\r\nunder increasingly dense operating conditions. Data mining\r\nbased tools and techniques are required to support and aid\r\noperators\u2019 (such as pilots, management, or policy makers)\r\noverall decision-making capacity. Within the NAS, the\r\nability to analyze fleetwide aircraft data autonomously is\r\nstill considered a significantly challenging task. For our\r\npurposes a fleet is defined as a group of aircraft sharing\r\ngenerally compatible parameter lists. Here, in this effort,\r\nwe aim at developing a system level analysis scheme. In this\r\npaper we address the capability for detection of fleetwide\r\nanomalies as they occur, which itself is an important\r\ninitiative toward the safety of the real-world flight operations.\r\nThe flight data recorders archive millions of data\r\npoints with valuable information on flights everyday. The\r\noperational parameters consist of both continuous and discrete\r\n(binary & categorical) data from several critical subsystems\r\nand numerous complex procedures. In this paper,\r\nwe discuss a system level anomaly detection approach based\r\non the theory of kernel learning to detect potential safety\r\nanomalies in a very large data base of commercial aircraft.\r\nWe also demonstrate that the proposed approach uncovers\r\nsome operationally significant events due to environmental,\r\nmechanical, and human factors issues in high dimensional,\r\nmultivariate Flight Operations Quality Assurance (FOQA)\r\ndata. We present the results of our detection algorithms on\r\nreal FOQA data from a regional carrier.", "distribution": [{"@type": "dcat:Distribution", "description": "chapter", "downloadURL": "https://c3.nasa.gov/dashlink/static/media/publication/2009_ICES_PF.pdf", "format": "PDF", "mediaType": "application/pdf", "title": "2009_ICES_PF.pdf"}], "identifier": "DASHLINK_424", "issued": "2011-07-05", "keyword": ["ames", "dashlink", "nasa"], "landingPage": "https://c3.nasa.gov/dashlink/resources/424/", "modified": "2025-03-31", "programCode": ["026:029"], "publisher": {"@type": "org:Organization", "name": "Dashlink"}, "title": "Fleet Level Anomaly Detection of Aviation Safety Data"}