FORECASTING TIME SERIES OF SMALL DURATION WITH THE USE OF ALGEBRAIC SEQUENCE AND INTERNAL SMOOTHING
The proposed forecasting method is based on the identification of the island algebraic sequence and finds a close to optimal balance between algebraic variability and smoothness of moving averages. The proposed method is compared with other forecasting methods (moving average, sequential smoothing, support vectors, ARIMA and composite methods). The proposed method of short-term time series forecasting is particularly well applicable for fault identification, service life forecasting and monitoring of technical objects. However, most of the dynamic processes occurring in contact systems can be described by short and rare events (the development of microcracks, electrical discharges between the contacting elements, etc.). Therefore, the development of reliable methods of time series prediction, in particular short series prediction, is a task of paramount importance and is successfully solved in this article.
Authors: M. V. Danilov
Direction: Informatics, Computer Technologies And Control
Keywords: Forecasting time series, core algebraic sequence, particle swarm method, internal smoothing
View full article