Title :
High dimension time series mining based on state transition chain analysis method
Author :
Lv Zhimin ; Zhang Kai ; Zhang Xiangwei ; Zong Shengyue
Author_Institution :
Nat. Eng. Res. Center for Adv. Rolling, Univ. of Sci. & Tech. Beijing, Beijing, China
Abstract :
In concerned with the high dimensions time series, which have the characteristic of multivariable, time-varying and time-lagging, were collected from multi-stage industrial processes, a method which is used to synchronize high-dimensions time series with temporal and spatial conversion is introduced in this paper. The high-dimensions time series is synchronized in spatial sampling by the conversion method. After the discretization for high-dimensions time series which is preprocessed by synchronization, first the control state is classified into normal state and high risk state by a simple association analysis; then using the method of state transition chain analysis, we successfully find the transition condition when the control system transform normal state into high risk state. This condition can be applied to reduce the quality defects of product and be used to guide the control strategy design of control system.
Keywords :
data mining; synchronisation; time series; association analysis; control state; spatial conversion; state transition chain analysis method; temporal conversion; time series mining; Algorithm design and analysis; Control systems; Data mining; Strips; Synchronization; Time series analysis; state space; state transitionl; temporal and spatial conversion; time series data mining;
Conference_Titel :
Fuzzy Systems and Knowledge Discovery (FSKD), 2011 Eighth International Conference on
Conference_Location :
Shanghai
Print_ISBN :
978-1-61284-180-9
DOI :
10.1109/FSKD.2011.6019814