Methodological Approach to Forming an Integral Anomaly Criterion Based on the Analysis of Time Series of Technological Parameters

The article considers a method for detecting anomalous modes in cyber-physical systems characterized by multidimensional time series, in which traditional anomaly detection methods lose sensitivity. An integral anomaly criterion is proposed, which includes components that account for the occurrence of anomalies both within a single input signal and in relation to other components of a multidimensional time series. The aim of the study is to determine the components of the anomaly criterion, the degree of significance of each component and its influence on the final result, as well as an approach to defining fixed boundaries for interpreting the criterion values.