Research On Complex Event Detection Based On Event Priorities In IOT For Manufacturing

In an environment like internet of things (IOT) for Manufacturing, the production site is extremely awful, a number of different sensors are deployed in all areas of the manufacturing site to monitor staff, materials, equipment, environment and other data on the worksite, huge amounts of data in the manufacturing industry have been generated. But, what users do share are only some specific semantic level events. How to quickly and efficiently extract the information of interest to the user from these data, which has been the hot issue in recent yearsIn order to get the useful information, the domestic and international scholars have proposed various solutions. Most of these schemes can only solved the problems for sequential flow of events, and all events can be kept in memory when composite event detection happen. But due to some reasons like the environmental interference, the delay of transmission network and so on, the data are disorder when reach the processor, with more and more sensors nodes, the data have exploded, and then all the complex events stored in memory are impossible to detect.According to the problems of abnormal events detection in manufacturing, this paper analyzes the characteristics of the field data in manufacturing and proposes a distributed complex event detection framework, combined with the scheme of distributed deployment of nodes. Aiming at detecting abnormal events occurred in the process of production quickly through the nodes deployed at the scene and the backend server. In order to define the abnormal events better, the degree of the seriousness of the events are dealt with, and how to apply the priority in complex event detection are put forward.Considering the different seriousness between events, we analyze the normal events, give a criteria for how to judge events’ priority, and design a efficient distributed event priority judging algorithm with an improved bloom filter. According to the priority of events which transfer to the server, a complex event detection algorithm based on events’ priority are proposed. Based on events’priority, with an effective replacement of the internal and external memory, the problem that the mass events can’t stored in the memory effectively can be solved effectively, and the high noise tolerance ensure the disorder event stream can do the correct complex event detection.This dissertation mainly researches complex event detection of IOT for Manufacturing, through analyzing the characteristics of data, proposes an event priority decision algorithm, and designs a priority complex event detection algorithm according to the priority of event. At last, compared with traditional complex event detection algorithm through simulation, we can see that this algorithm is more efficiency in IOT for Manufacturing, and the algorithm has a larger increase in time relative.