The identification of data anomalies from information sensors based on the estimation of the correlation dimension of the time series attractor in situational management systems

Abstract

Purpose: The goal is the timely detection of uncharacteristic behavior of the observed processes in the systems of situational management, leading to the development or occurrence of emergency situations. Methodological approach: In the article, it is proposed to analyze the change dynamics in the correlation dimension of the attractor in order to detect anomalies in the behavior of the observed process. A sharp change in the correlation dimension is a reflection of the uncharacteristic (anomalous) nature of the data of the observed processes. This anomaly is a consequence of external influences on the generating system and requires an analysis of the causes of its occurrence. Uniqueness/value: The uniqueness of the proposed approach consists in the fact that an abrupt change in the correlation dimension of the attractor is the information about the occurrence of uncharacteristic behavior of the observed system. The value of the study is determined by the relevance of the problem of modeling the development and occurrence of emergency situations in situational management systems based on the analysis of the time series of observed processes. Summary: The proposed approach is designed to identify the critical states of the generating dynamical systems by their time series. Timely response to the transition of the monitored system to a critical state will allow preventing any critical consequence