METHOD AND APPRATUS FOR DETECTING ANOMALY OF VEHICLE BASED ON EUCLIDEAN DISTANCE MEASURE

摘要

A vehicle abnormality detection method based on Euclidean distance measurement technique is disclosed. The Euclidean distance measurement technique-based symptom detection method is performed in a vehicle abnormality detection device, and includes collecting vehicle state data for n (n is a natural number of 1 or more) vehicles in a certain interval, Generating filtered vehicle state data by performing filtering on data having no fluctuation in data values among the vehicle state data; calculating a m-dimensional reduced magnitude m by factor analysis of the filtered vehicle state data, Extracting a detailed factor of each of the detailed factors and generating state data for each of the detailed factors, extracting a reference value for vehicle anomaly detection using the traffic state information, with the condition data for calculating a Euclidean distance (also known as D _k, the "first distance"), the Calculating an Euclidian distance (hereinafter, also referred to as a second distance) between the reference value and the state data for each of the sub-factors of the vehicle to be analyzed; Detecting an abnormality for each of the sub-factors using the second distance, and providing an anomaly detection notification including information about a cause of the vehicle abnormality when the vehicle abnormality is detected.