Research On Construction And Matching Method Of Streaming Big Data Event Template Based On B-Spline Curve

Streaming data is a contiguous sequence of data.Features such as fast,large,and continuous arrival in time that cannot be processed in the usual way.In recent years,along with China's economic development,the application of streaming data has increased,such as sensor networks,log systems,real-time monitoring and so on.These devices or programs generate a large amount of real-time data in their real production environment,which is a valuable data resource for production and life.Due to its large number of real-time features,streaming data cannot be processed in a general way,and large data processing techniques are required.In addition,there is an important type of streaming data system,which usually has some kind of events,such as earthquake monitoring systems,coal mining and mining monitoring systems.Such systems typically monitor events occurring on streaming data to guide production and life through monitored events.In the process of coal mining,multiple sensor base stations are placed around the mine to monitor the microseismic signals generated during coal mining.The microseismic signal will experience energy attenuation and interference during the propagation of air through the rock.Therefore,for different sensors,the monitoring of the same event is not the same in data representation;the same sensor may have different manifestations for multiple monitoring of the same event.This results in the same event behave differently in horizontal time and in the longitudinal direction.Therefore,this paper studies the classification of events on streaming data.Firstly,according to the characteristics of streaming data,this paper gives the event definition and event template definition on streaming data.On this basis,the average situation of the event is obtained,and then the event with the smallest average event error is calculated as the basic scale event according to the cosine theorem.Based on the event,the normalized processing method of the streaming data event based on linear transformation is given.All the remaining events of the same type are transformed with reference events as basic reference events,so that similar events of different sizes in the event domain and the energy domain are organized into the same range.The data points at the same location are then clustered using the corresponding clustering method,thus obtaining a B-Spline input curve.The control vertices of the B-Spline curve are obtained by an improved genetic algorithm,and then the event template is constructed.Secondly,The start and end points of the streaming data event are obtained by the time interval method.And use the piecewise cumulative approximation method to perform data compression processing on streaming data events to avoid excessive calculation.The data is then transferred to the location of the event template,and the event is matched using the distributed environment to calculate the match between the event and each template.Finally,the experimental analysis of the streamlined big data event template classification method based on B-Spline curve is carried out,and the template generation efficiency analysis is carried out by comparing experiments with different methods and changing different numbers of events.Experiments show that the proposed B-Spline curve-based streaming big data event template classification method has higher execution efficiency and lower efficiency in resource utilization than similar methods.