Methods for Forecasting Critical States of the Technological Process in the Evolutionary Management of Continuous Multi-Stage Production

The article deals with the problem of detecting anomalies in complex dynamical systems under conditions of partial uncertainty inherent in continuous multistage productions. Possible approaches to solving the problem of early warning of critical states of the technological process are considered based on the data of continuous monitoring and analysis of the values of diagnostic indicators received during the operation of technological equipment from various sensors. The state of the system is estimated based on short time series formed by sensor readings. Thus, each sensor registers the values of a random process, the characteristics of which are unknown a priori. A set of sensors registers individual values of a multidimensional random process with unknown characteristics. Proposed methods for detecting anomalies in a multi-sensor distributed system allow us to combine the results of measurements obtained from individual sensors. This is necessary in order to draw conclusions about the possible transition of the technological process to a critical state, taking into account the correlation of data in time series.