Study on algorithm of dependent pattern discovery of multiple time series data stream

For the question of traditional MSDD algorithm to discovery useful structure model from time series that consisted of multiple data streams can not pruning the nodes well, and also can not express the time relationship of model intuitively, through the research, we put forward a kind of algorithm which can discover time series dependent pattern structure based on multiple data streams: the algorithm of time window move screening(TWMA).We adopted the strategy of the sequence of events to discover dependent pattern from more flow time series, compared with MSDD, our method is more intuitively in express and more flexible in process of patterns discovery.