Detection of Violations of Technological Process Dynamics By Structural Analysis Method

Abstract

An algorithm is proposed for analyzing non-stationary time series obtained by monitoring technological processes or properties of the developed unidimensional fibre products (yarn, sliver, roving, tow, complex yarn, etc.). The algorithm is based on the correlation, known from the theory of chaos dynamics, between the trend and the random variations of values and consists in simultaneous evaluation of the trend of average and a range of variations. It can be used to identify and predict the type of nonstationary state in terms of the mean level and variations. The information gathered with its aid will make it possible to predict possible structural changes (bifurcation) in the characteristic features of the trends of time series. The algorithm can be used in many areas of analysis, control, and prediction of technological and technoeconomic processes.