Research On Complex Event Detection Method For Multivariate Time Series Data

With the continuous development of information technology,many areas of society have established more and more business application systems,a lot of data were generated by different systems.With the huge amount of data,it is increasingly meaningful to combine different business system data and get the potential values of data.It is complex event detection technology that abstracting the business data as an event sequence,and describing the potential value of the composite data as a specific event matching structure through the event description method.And then the event detection engine found the matching structure of the event sequence from a large number of event flow,and finally output the data fusion results.However,in the traditional event description,the input event flow of the event engine is a single atomic event type.The event predicate constraint contains simple attribute value comparison operation and simple aggregation operation.The time constraint between events can only be "after","before" and so on.This makes the traditional detection method can not meet the requirements such as medical,financial and other relatively accurate time requirements,event predicate constraints require more complex applications.Thus,a complex event detection method that can support multiple event inputs,support inter-event quantitative timing constraints,and complex predicate constraints is very necessary.In order to realize complex event detection,which supports multi-event input,support inter-event quantitative timing constraints and complex predicate constraints,a complex event detection method towards multi-temporal data was proposed in this paper that based on traditional complex detection model,and defined TCN-based timing constraints event time representation method and predicate constraint representation method which based on time feature constraint.At the same time,this paper enhances the event input mode and supports multiple event input,and proposes the event detection algorithm and the parallel detection method,which makes the complex event detection method more efficient.The main contributions of this paper are as follows:(1)Event definition: Before the event description,it is important that defines the concept of events,so the concept of the event is re-defined in this paper,which contain the atomic events,aggregation events and complex events.(2)Event description method: In this paper,the TCN time constraint network is introduced to represent the representation of inter-event quantization timing,and the time feature representation method is introduced to express the complex predicate constraint,and the aggregation event description model and complex event description model are proposed.(3)The basic complex event detection algorithm proposed: A algorithm which Based on the traditional complex event detection system SASE is proposed to support the quantitative timing constraints and complex predicate constraints,and it supports the multiple events flow input.(4)Parallel complex event detection algorithm proposed: Some performance bottlenecks exist in the basic detection algorithm,so a parallel complex event detection method based on event stream slicing is presented in this paper.In order to verify the validity of the method which proposed in this paper,we develop a complex event detection prototype system and design a number of experiments,which are verified experimentally from the aspects of processing ability,the correctness of event detection,data size,length of matching sequence,sliding window size and so on.The experimental results show that the proposed method can effectively support the complex event detection of multiple event input,quantization timing constraint and complex predicate constraint,and has high efficiency.