KR-20260063754-A - Method and system for generating cloud-level lost data for vehicle data abnormality detection technology

KR20260063754AKR 20260063754 AKR20260063754 AKR 20260063754AKR-20260063754-A

Abstract

A cloud-level lost data generation method according to one embodiment of the present invention comprises: a step of classifying sensor data generated inside a vehicle by sensor type and storing it in a data archive; a step of searching for lost data among the sensor data stored in the data archive using a DTW algorithm; and a step of performing sampling of the lost data from the data archive based on the similarity rank of the searched lost data. By doing so, data in a situation most similar to the lost data can be extracted based on the history of the vehicle data.

Inventors

장수현
신대교
장준혁
장성현
안병만
노용철

Assignees

한국전자기술연구원

Dates

Publication Date: 20260507
Application Date: 20241031

Claims (12)

The system acquires sensor data generated inside the vehicle; The system classifies acquired sensor data by sensor type and stores it in a data archive; The system utilizes a Dynamic Time Warping (DTW) algorithm to search for lost data among sensor data stored in a data archive; and A method for generating cloud-level lost data for vehicle data anomaly detection technology, comprising the step of the system performing sampling of lost data from a data archive based on the similarity rank of the discovered lost data.
In claim 1, The step of storing in the data archive is, The system performs clustering by sensor type on acquired sensor data; and A method for generating cloud-level lost data for vehicle data abnormality detection technology, characterized by including the step of the system storing sensor data classified by cluster in a data archive.
In claim 2, The step of performing clustering by sensor type is, Clustering is performed based on the similarity of the sensors, and The step of performing sampling is, A cloud-level lost data generation method for vehicle data anomaly detection technology, characterized by extracting data characteristics within a cluster of the discovered lost data when the lost data is discovered.
In claim 3, The step of performing sampling is, A cloud-level lost data generation method for vehicle data abnormality detection technology, characterized by comparing, when lost data is detected, first sensor data acquired at a time prior to the acquisition of the detected lost data within the same cluster with second sensor data acquired at a time after the acquisition of the detected lost data, based on the acquisition time of the detected lost data.
In claim 4, The step of performing sampling is, A cloud-level lost data generation method for vehicle data abnormal state detection technology, characterized by extracting multiple similar sections by utilizing time-series data stored in a data archive and a DTW algorithm when comparing first sensor data and second sensor data.
In claim 5, The step of performing sampling is, A cloud-level lost data generation method for vehicle data abnormal state detection technology, characterized by calculating similarity between clustered sensor data based on the extracted multiple similar sections when multiple similar sections are extracted.
In claim 6, The step of performing sampling is, A cloud-level lost data generation method for vehicle data abnormal state detection technology, characterized by generating a Rank List based on the similarity between clustered sensor data when similarity between clustered sensor data is calculated.
In claim 7, The step of performing sampling is, A cloud-level lost data generation method for vehicle data anomaly detection technology, characterized by generating a Rank List based on similarity between sensor data, and performing sampling of lost data from a data archive based on a Sliding Window at the top of the generated Rank List.
In claim 1, The step of storing in the data archive is, A cloud-level lost data generation method for vehicle data abnormal state detection technology, characterized by classifying acquired sensor data by sensor type and storing it in a FIFO (First In First Out) manner.
A communication unit for acquiring sensor data generated inside the vehicle; and A cloud-level lost data generation system for vehicle data anomaly detection technology, comprising: a processor that performs clustering by sensor type on acquired sensor data, stores sensor data classified by cluster in a data archive, searches for lost data among sensor data stored in the data archive using a Dynamic Time Warping (DTW) algorithm, and performs sampling of lost data from the data archive based on the similarity rank of the searched lost data.
A step in which the system performs clustering by sensor type on sensor data generated inside the vehicle; The system stores sensor data classified by cluster in a data archive; The system utilizes a Dynamic Time Warping (DTW) algorithm to search for lost data among sensor data stored in a data archive; and A method for generating cloud-level lost data for vehicle data anomaly detection technology, comprising the step of the system performing sampling of lost data from a data archive based on the similarity rank of the discovered lost data.
A clustering execution unit that performs clustering by sensor type on sensor data generated inside a vehicle and stores the sensor data classified by cluster in a data archive; A storage unit including a data archive in which sensor data classified by cluster is stored; A lost data search unit that searches for lost data among sensor data stored in a data archive using a DTW (Dynamic Time Warping) algorithm; and A cloud-level lost data generation system for vehicle data anomaly detection technology, comprising: a sampling execution unit that performs sampling of lost data from a data archive based on the similarity rank of the discovered lost data.

Description

Method and system for generating cloud-level lost data for vehicle data abnormality detection technology The present invention relates to a method and system for generating lost data, and more specifically, to a method and system for generating lost data by utilizing machine learning-based clustering and dynamic time warping techniques. Lost data refers to cases where the original data cannot be preserved and is lost due to viruses, hardware damage, communication failures, etc. Conventional methods for generating lost data commonly use interpolation and replacement with average values to process the lost data. However, these conventional methods for generating lost data have difficulty reflecting the changing state of the data, and in particular, when interpolation is applied after the occurrence of abnormal data with extreme values, a problem may arise where the entire interpolated data is applied as an outlier. To overcome these problems, an archive sampling method has been proposed that preserves the original data for abnormal and normal data by storing and updating data in an archive and sampling from the data archive when missing data occurs; however, since sensor data with non-periodic characteristics has limitations in sampling, it is necessary to explore solutions for this. FIG. 1 is a drawing provided in the description of the configuration of a cloud-level lost data generation system for a vehicle data abnormal state detection technology according to an embodiment of the present invention. FIG. 2 is a drawing provided for a more detailed configuration description of the processor illustrated in FIG. 1. FIG. 3 is a flowchart provided in the description of a cloud-level lost data generation method for a vehicle data abnormal state detection technology according to an embodiment of the present invention. FIG. 4 is a diagram provided to explain the process of clustered sensor data being stored in a data archive through a cloud-level lost data generation system for vehicle data abnormal state detection technology according to an embodiment of the present invention. FIG. 5 is a drawing provided to explain the process of searching for lost data through a cloud-level lost data generation system for vehicle data abnormal state detection technology according to an embodiment of the present invention. FIG. 6 is a drawing provided to explain the process of extracting a plurality of similar sections by utilizing time-series data stored in a data archive and a DTW algorithm through a cloud-level lost data generation system for vehicle data abnormal state detection technology according to an embodiment of the present invention. FIG. 7 is a drawing provided to explain the process of calculating similarity between clustered sensor data based on a plurality of similar sections extracted through a cloud-level lost data generation system for vehicle data abnormal state detection technology according to an embodiment of the present invention. FIG. 8 is a drawing provided for explaining the process of generating a Rank List based on similarity between sensor data through a cloud-level lost data generation system for vehicle data abnormal state detection technology according to an embodiment of the present invention, and FIG. 9 is a diagram provided to explain the process of performing sampling of lost data in a data archive using a Rank List generated through a cloud-level lost data generation system for vehicle data abnormal state detection technology according to one embodiment of the present invention. The present invention will be described in more detail below with reference to the drawings. To clearly explain the invention, parts unrelated to the description have been omitted from the drawings, and in the drawings, the width, length, thickness, etc., of the components may be exaggerated for convenience. FIG. 1 is a diagram provided in the configuration description of a cloud-level lost data generation system for vehicle data abnormal state detection technology according to one embodiment of the present invention. The cloud-level lost data generation system for vehicle data abnormal state detection technology according to the present embodiment (hereinafter collectively referred to as the 'system') extracts data in a situation most similar to the lost data based on the history of vehicle data, and the abnormal state detection algorithm can accurately detect the abnormal state. To this end, the system may include a communication unit (100), a processor (200), and a storage unit (300). The communication unit (100) is equipped with a communication module connected to a network, so that it can acquire sensor data generated inside the vehicle. The storage unit (300) is provided to store programs and data necessary for the operation of the processor (200). For example, the storage unit (300) may include a data archive in which sensor data classified by cluster is stored. The processor (200) can extract data in a situati