ABNORMALITY DETECTING DEVICE AND ABNORMALITY DETECTING METHOD

To enable, in an abnormality detection based on a plurality of time-series sensor signals, a fast-speed process together with an abnormality measure calculation and clustering while an abnormality detection sensitivity is maintained.SOLUTION: An abnormality detecting device adjusts a certain number...

Full description

Saved in:
Bibliographic Details
Main Authors MIYABE SHOZO, KASHIWA TAKUKI, SHIBUYA HISAE, NODA TOJIRO
Format Patent
LanguageEnglish
Japanese
Published 14.01.2021
Subjects
CONTROL OR REGULATING SYSTEMS IN GENERAL
CONTROLLING
FUNCTIONAL ELEMENTS OF SUCH SYSTEMS
MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS ORELEMENTS
PHYSICS
REGULATING
Online AccessGet full text

Cover

More Information
Summary:To enable, in an abnormality detection based on a plurality of time-series sensor signals, a fast-speed process together with an abnormality measure calculation and clustering while an abnormality detection sensitivity is maintained.SOLUTION: An abnormality detecting device adjusts a certain number of feature vectors belonging to respective clusters by performing clustering on the feature vectors within a specified learning time period, selects the one cluster according to the newly extracted feature vector, and measures an abnormality measure on the basis of a reference vector that is calculated using all the feature vectors belonging to the selected cluster.SELECTED DRAWING: Figure 1 【課題】複数の時系列センサ信号に基づく異常検知において、異常検知感度を維持しながら、異常測度算出、クラスタリングともに高速に処理可能にする。【解決手段】異常検知装置において、指定された学習期間の特徴ベクトルをクラスタリングして各クラスタに属する特徴ベクトルを一定数に調整し、新たに抽出した特徴ベクトルに応じて１個のクラスタを選択し、選択したクラスタに属する全特徴ベクトルを用いて算出される基準ベクトルに基づいて異常測度を算出する。【選択図】図１
Bibliography:Application Number: JP20190117357